Project Summary Substance use disorders (SUDs) are increasing in prevalence and are already a leading cause of disability, due in part to the fact that our understanding of the underlying pathophysiology is incomplete. Like most neuropsychiatric syndromes, SUDs are highly heterogeneous, and distinct mechanisms may be operative in some individuals but not in others, even within a single diagnostic category. Furthermore, SUDs frequently co- occur with depression, anxiety, and other psychiatric syndromes, complicating efforts to identify molecular and circuit-level mechanisms, and disentangle them from those involved in mood and anxiety disorders. Diagnostic heterogeneity is thus a fundamental obstacle to developing better treatments, identifying biomarkers for quantifying risk for different forms of addiction, and predicting treatment response and relapse. Recently, we developed and validated an approach to discovering and diagnosing subtypes of depression using fMRI measures of functional connectivity, which in turn predicted subtype-specific clinical symptom profiles and treatment outcomes. Here, in response to PAR-18-062, we propose a secondary data analysis that would extend this approach to SUDs, leveraging multiple deeply characterized and large-scale neuroimaging datasets. Our central hypothesis is that individual differences in mechanisms underlying impairments in response inhibition and salience attribution (iRISA) are mediated by distinct forms of dysfunctional connectivity in addiction-related circuits, which in turn interact and give rise to distinct neurophysiological addiction subtypes. In Aim 1, we will use statistical clustering and machine learning methods to delineate these subtypes and optimize classifiers (fMRI biomarkers) for diagnosing them in individual patients, focusing initially on cocaine addiction. In Aim 2, we will validate these subtype-specific biomarkers by first replicating them in a new dataset and then evaluating their longitudinal stability and predictive utility. In Aim 3, we will test whether subtype-specific circuit mechanisms generalize to mediate iRISA functions in other forms of addiction, and define their interactions with distinct mechanisms mediating anhedonia and anxious arousal in patients with comorbid depression and anxiety.